Figure 1

Two-level approach for structural ensemble analysis. Diagram of the proposed approach for structural ensemble analysis, including the optimization of the Self-Organizing Map performed in this work. Optimization protocol: the Experimental Design allows the training of SOMs with different parameters (SOMs trained with different parameters); the output of the SOMs training is used to optimize the SOM parameters (Design analysis and SOM parameters optimization) and to find the Optimal SOM which is trained with different sampling rates (SOMs trained with different sampling rates); the output of this training process is submitted to a Chi-square test to confirm the Optimal SOM. Two-level approach: the prototype vectors of the optimal SOM trained with the MD conformational ensemble are submitted to a clustering process (Cluster analysis) which extracts the Representative conformations.